Uncertainty estimation in the analysis of pesticide residues in olive oil using data from proficiency tests

Abstract

In this work we report the results for estimating the measurement uncertainty (MU) following up the application of two different approaches, relatively the top-down procedure, by using proficiency test data. We have focused the estimation on the olive oil matrix. We used the analytical data obtained from five selected editions of the Proficiency Tests (PTs, from 2007 to 2011) on pesticide residues in olive oil to estimate the MU. These PTs have been organized by Istituto Superiore di Sanità annually in cooperation with International Olive Council (IOC) since 1997. The number of participants in each trial ranged from 10 to 43. We used a total of 34 pesticide results. The expanded uncertainty U (c) was calculated using a covering factor k = 2 for a confidence interval of 95%. In the approach 1, the within–laboratory reproducibility standard deviation is combined with estimates of the method and laboratory bias using PTs data. In the approach 2, the way of estimating the MU is based only on the bias that the laboratory has obtained participating in a sufficient number of the IOC proficiency tests. Comparing the relative expanded uncertainty based on these different approaches we notice values quite constant and close, from 42% to 48%. Moreover, these calculated expanded uncertainties are less than the default value of 50% (corresponding to a 95% confidence level), adopted from European guidance document SANCO based on the fit-for-purpose relative standard deviation (FFP-RSD).